Windmill

Windmill Launch Week — That's a wrap 🎉                                                                              Five days, five major releases. Here's everything we shipped: ➡️ Day 1 — Full-code apps                                      Build frontends in React or Svelte directly on Windmill. Your own npm packages, hooks, routing, no low-code constraints. With built-in AI skills so Claude Code and Codex understand your entire stack.  ➡️ Day 2 — Data Tables & Ducklake                                  A built-in PostgreSQL database and an S3-backed lakehouse, right inside Windmill. Your data no longer needs to live outside your automation platform.                                ➡️ Day 3 — AI Sandboxes Run Claude Code, Codex, or any coding agent in isolated Windmill jobs. Scoped permissions, resource limits, persistent volumes, full audit logs. Production-grade agent infrastructure.      ➡️ Day 4 — Git Sync & Workspace Forks                                Your entire workspace is now version-controlled. Branch it, test in isolation, merge when ready. Enterprise collaboration, built in.                                                   ➡️ Day 5 — Workflow-as-code Define complex workflows in pure TypeScript or Python. Checkpointing, replay, fault tolerance, no YAML, no DSL.          The common thread: Windmill is becoming the single platform where you build, run, and manage all your internal apps. Backend logic, frontends, data, AI agents, and orchestration. 📖 Read all the full recap: https://lnkd.in/d9MNgkHj

Launch Week, final Day ! 🎉 Introducing Workflow-as-code Windmill already had a powerful visual flow editor. But some workflows are just easier to write in code. Windmill now lets you define complex workflows in pure TypeScript or Python. No YAML, no DSL, no new infrastructure. ➡️ Gives you the full power of your language while Windmill handles checkpointing, replay, and fault tolerance. ➡️ Zero wasted resources. Workflows suspend and release workers while waiting for sleep, approval, or child jobs. ➡️ Build full workflows in your IDE with the Windmill CLI and built-in AI skills for Claude Code and Codex. 📖 Read full blog post: https://lnkd.in/dNdj5EWJ

Launch Week, Day 4 : 🔆 Introducing Git sync & workspace forks Most automation platforms treat version control as an afterthought. Every change is a production deploy. No staging, no review, no rollback. That doesn't work when multiple teams share the same platform and downtime has real consequences. With Windmill, your entire workspace is now version-controlled. Every script, flow, app, resource, and variable. Not just files in a repo, but the full state of your platform, tracked with the same Git workflows your engineering teams already use. ➡️ Git sync automatically commits every deploy to your repository. ➡️ Workspace forks let you branch your entire workspace, develop and test in isolation, and merge back when ready. This is what enterprise-grade collaboration looks like. Multiple teams working in parallel, staging environments that mirror production exactly, and a full audit trail of every change. 📖 Blog post: https://lnkd.in/dPKDAU2i

Launch Week, Day 3: 🚀 Introducing AI Sandboxes We see more and more agents running autonomously on machines. Full disk access, your keys, no audit trail. While it works for testing and prototyping, it becomes dangerous when running at scale in enterprise environments. Agents in production need isolation, scoped permissions, resource controls, and a log of everything they did. Windmill already built this entire infrastructure for human operators. Now we're making it available for AI agents. Windmill now runs Claude Code, Codex, or any coding agent in isolated sandboxes with scoped access. ➡️ Each agent run is a Windmill job: scoped permissions, resource limits, full audit logs. ➡️ Persistent volumes let agents pick up where they left off between runs. ➡️ Orchestrate agents with flows: chain runs, add human approvals, run steps in parallel. 📖 Read full blog post: https://lnkd.in/dSgPWmjZ

Launch Week, Day 2: 📊 Introducing Data Tables & Ducklake In Windmill, you can already build and orchestrate all your internal apps. The one thing still living outside was the data itself. Now your scripts and flows can read and write data without leaving Windmill. ➡️ Data Tables: a built-in PostgreSQL database scoped to your workspace. Your scripts and flows can read and write to it directly, no external database to provision. ➡️ Ducklake: an S3-backed lakehouse for larger datasets. Store millions of rows, query them with SQL, all managed from Windmill. Both browsable through a built-in Database Studio. Your entire pipeline, from ingestion to transformation to the app that displays it, lives in one place. 📖 Read full blog post: https://lnkd.in/dvSvtSW6

Launch Week, Day 1: 🚀 Introducing full-code apps Until now, Windmill had a low-code app builder to build frontends on top of your scripts and flows. Great for dashboards and forms, but limiting when you need custom interactions or your own component libraries. Now you can build full-code apps with React and Svelte, directly on Windmill. ➡️ Use React or Svelte with your own npm packages, hooks, routing, and component libraries. No low-code constraints. ➡️ Use your Windmill scripts and flows for backend. Windmill auto-generates a typed API so your frontend calls backend with full type safety. ➡️ Built-in AI assisted development: the Windmill CLI now generates skills so Claude Code and Codex understand your project structure, backend runnables, and data schema out of the box. You can now build full-stack apps in local dev with AI that knows your entire stack. 📖 Read full blog post: https://lnkd.in/dmCWevrb

Launch Week #2 is coming. March 30 to April 3.                                                                                      Last year we shipped a full workflow engine. This time we're going deeper into building the best orchestration platform for the age of AI.                                                          5 days. 5 major releases. From data management to AI sandboxes, from full code frontends to new deployment workflows. One feature a day, all week long.                           Follow along at https://www.windmill.dev/ 

For enterprise sales teams, consistent demo environments are critical. But when your platform runs real-time APIs across meetings, phone, chat, and contact center, you can't just spin up a sandbox.                                   Zoom's Global Architecture organization needed to support 400+ Solution Engineers with high-quality demos, without each SE building their own environment manually.                                              They built a centralized demo platform on Windmill. When an SE schedules a demo for Tuesday, the platform kicks off hydration on Friday. Over the weekend, scheduled jobs deliver chat messages, voicemails, and documents at realistic intervals. By demo day, the account looks like it's been actively used for weeks. Today, 60+ Global Architects contribute to the platform, and 400+ solution engineers get realistic, ready-to-go environments for every customer conversation. 🔗 https://lnkd.in/dU5fhyeN Thank you Frederick Loucks, Tom Pacyk and the Zoom team for your collaboration on this case study.

For managed service providers, scaling automation across hundreds of customer environments creates a licensing dilemma.                                                                                                                                                                                             Most orchestration platforms use rigid per-user or per-device pricing. When you are onboarding 150 enterprise customers incrementally, committing to massive upfront costs before proving value isn't viable.                                                                                                                                                   Axians CS Netherlands, an IT infrastructure provider serving government agencies, hospitals, and large enterprises, needed multi-tenant orchestration with strict customer isolation, but with costs that grow proportionally to adoption.                                                                                                                                   They evaluated more than 10 platforms before choosing Windmill. With isolated workspaces per customer and usage-based pricing, they now have a centralized platform where everything is audited, registered, and logged. 🔗 https://lnkd.in/dUKqxbTR Thank you Ben Willems and the Axians team for your collaboration on this case study.

In January 2024, Airplane.dev announced it would shut down, forcing teams to find and migrate to alternative platforms. Panther, a cybersecurity SIEM platform, saw an opportunity to rethink their internal tooling architecture. They chose Windmill for its open-source foundation and built a platform-agnostic architecture where business logic lives in a separate Python library. If they ever need to migrate again, they rewrite the thin orchestration layer, not their entire platform. Today, 60% of their internal operations runs on Windmill, described as "among the most stable components of our infrastructure." 🔗 https://lnkd.in/eeAus9bw Thank you BJ Maldonado, WIDB and the Panther team for your collaboration on this case study.